All five weekday entries (Monday August 31 through Friday September 4) are present in the research log. Three items are deduplicated across weeks rather than within this one and are reported as movement rather than as new findings: the Itron / T&D World “DERMS alongside SCADA” essay (same June 9 URL) appeared in last week’s digest and is carried here only for the procurement-specification detail the September 4 entry foregrounds; CalChoice’s Lunar Energy DERMS was reported in the August 28 digest as a targeted launch and appears here only because the target date arrived; and the New Jersey BPU virtual-power-plant straw proposal, covered August 28, is carried only as context for the compensation-design flag. Xcel Minnesota’s Capacity*Connect program is a standing watch item in this series and is treated here as a cost benchmark rather than as news. Several items carry pre-week publication dates — the FERC RM26-4 show-cause orders from June 18, the Duke Nicholas Institute flexibility economics from June 26, the DOE generation-side 202(c) analysis from June 23, the SEPA/DSIRE Q2 policy roundup from July 23, the FERC statistical-sampling order issued July 29 and reported August 13, the Regulatory Assistance Project webinar synthesis from August 19 — and they are reported on the day they entered the log with original dates shown. Two verification items this series has carried for four weeks resolve this week; both are noted where they land.
Last week’s digest argued that the same megawatt of large-load flexibility had become available two ways — negotiated or commandeered — and that the only strategic variable left was which instrument a utility ended up holding. This week the sector did something more revealing: it priced the same megawatt three different ways inside five days, and a federal regulator removed the measurement rule that had been keeping a fourth version off the books entirely.
The compelled version is free and arrived first. On September 1, in response to a same-day application, the Department of Energy issued Section 202(c) Order No. 202-26-41 to PJM — the third invocation of 1935 wartime authority in 2026 — directing specified units to run and authorizing PJM, with its transmission owners and electric distribution companies, to switch backup generation at large loads onto their own generators as a last resort before or during an Energy Emergency Alert 3, across the 13-state footprint. The load class in scope is the familiar one: 50 MW or greater at a single delivery point, which is to say hyperscale data centers. Nothing is paid, nothing is metered for settlement, and no reusable capability is left behind.
The rate-based version costs $2,150 per kW. Minnesota’s PUC approved Xcel’s Capacity*Connect — up to $430 million for up to 200 MW of distribution-sited batteries in 1–3 MW increments — over clean-energy objections that it should have been opened to independent developers. At roughly $2,150/kW it runs more than triple the $624/kW of Xcel’s own 125 MW Colorado Advanced VPP, which is the entire substance of the competitive-procurement fight.
The third-party-funded version costs ratepayers nothing. On September 3, PG&E, Google and Rewiring America launched SHARE — Smart Home Assets for Reliability and Efficiency — enrolling roughly 21,000 already-installed Tesla and Sunrun batteries, Renew Home thermostats and EVs in Santa Clara and Alameda counties, deploying new Carrier battery-enabled heat pumps alongside them, and aggregating the fleet with Demand Side Analytics to relieve regional transmission constraints as early as fall 2026. The entire proof-of-concept is funded by Google, not recovered through rates.
And then the measurement gate came down. FERC, ruling for Voltus and the Mission:data Coalition in Docket EL26-4, found it “unjust and unreasonable” for PJM to require interval smart-meter data that aggregators cannot obtain, and held that statistical sampling “can be as or more accurate” across large numbers of homes. The Commission’s own finding on what the status quo cost: PJM was losing at least 4.9 GW of residential demand-response and virtual-power-plant capacity “at a time when PJM is experiencing unprecedented load growth driven in part by hyperscale data center development, threatening reliability.”
Put the four together and the week’s lesson is not about technology at all. A utility can obtain demand-side capacity for free by federal directive and get nothing durable; for $2,150/kW on the rate base and own it outright; for nothing from a corporate counterparty and own the operational learning but not the asset; or for the cost of an M&V rule change and unlock gigawatts that were already sitting installed in customers’ homes. Only the last of those was free and additive, and it took a Federal Power Act proceeding to get it.
Underneath, the rails kept going in. The Congressional Research Service issued a federal reference document quantifying what flexibility already buys — NERC found the summer 2026 ERCOT peak forecast runs 3.7 GW (4.6%) lower than 2025 because more data centers can be curtailed. MISO filed a reliability framework at FERC that would make visibility, phasor measurement, ramp limits and ride-through preconditions of interconnection for “computational loads” at 25 MW of IT demand. SEPA counted 25 new large-load tariffs proposed in 2026 with roughly one-quarter now carrying a dispatchable-flexibility option. PJM’s market monitor put a number on the bill those tariffs are trying to allocate: data-center load accounted for 9% ($10.48/MWh) of wholesale power cost through July and $29.4 billion of capacity-revenue increases across four auctions.
The through-line: last week flexibility acquired a price and a procurement channel. This week it acquired a funding structure question — and the answers on the table differ by more than a factor of three.
🔋 Energy Storage
President Trump’s August 26 executive order restricting foreign-made bulk-power components threatens the hardware layer beneath every storage-anchored VPP in this series, and BloombergNEF expects delays and outright cancellations while developers wait for guidance. The order blocks certain foreign-produced batteries, transformers, inverters and related digital products from connecting to the U.S. grid after its August 26 effective date, and gives the Energy Secretary discretion over equipment already installed — a retroactive exposure that is unusual and that developers cannot price. It layers onto the Treasury’s Foreign Entity of Concern rules finalized in February 2026 under the One Big Beautiful Bill Act, which condition federal tax credits on limited Chinese supply-chain reliance. The concentration figures are the reason this is a first-order risk rather than a paperwork problem: China accounts for roughly 80% of global lithium-ion battery supply-chain capacity and supplied about 40% of U.S. inverter volumes in 2025. BNEF’s expectation is deployment delays as developers seek alternative suppliers or redesign projects outright.
For the avoided-cost frame this series runs, the order is the counterweight to everything else in this week’s digest. If battery and inverter costs rise or schedules slip, the economics tighten on precisely the programs the regulatory momentum is creating — New Jersey’s behind-the-meter solicitation, Xcel’s Advanced VPP, California’s DSGS storage tracks, and the 200 MW Xcel Minnesota is about to procure at $2,150/kW. The resource that is not exposed is the one this series keeps benchmarking: demand response and energy efficiency at roughly $20.70/MWh have no import content at all. A portfolio weighted toward thermostats, water heating, EV managed charging and load flexibility carries supply-chain risk that a battery-heavy portfolio does not, and that is a procurement argument that did not exist two weeks ago.
For DERMS specification the implication is about breadth rather than depth: an orchestration platform whose value case rests on batteries is exposed to a policy shock it cannot hedge, while one that can dynamically reweight thermostats, EVs and load curtailment against constrained battery supply gains option value. BNEF notes the longer-term hit may be cushioned by new domestic capacity, with four U.S. battery plants online this year and more expected by year-end. That is a two-to-four-year cushion arriving after a procurement cycle that is happening now. (Source: Utility Dive, September 1 — September 3 entry)
Carried benchmark — Xcel Minnesota Capacity*Connect at ~$2,150/kW. This series has tracked the program since approval; the September 2 entry supplies the arithmetic that makes it useful as a comparable. The Minnesota PUC authorized up to $430 million for up to 200 MW of distribution-sited batteries in 1–3 MW increments over two years, dispatched by Xcel as a utility-operated virtual power plant close to load. Clean-energy intervenors and Vote Solar argued the program should have been opened to independent developers; the implied ~$2,150/kW against $624/kW for Xcel’s own 125 MW Colorado Advanced VPP is the spread that argument rests on. The procurement point worth extracting is orthogonal to the ownership fight: a fleet of hundreds of small distributed batteries under a single owner still needs the same resource-formation, dispatch and settlement layer a third-party aggregation needs, so the platform decision does not resolve itself by choosing an ownership model. The standing watch item is unchanged — the November 2027 Integrated Grid Plan avoided and deferred distribution value estimates, with quarterly reports before then, are what will eventually test whether $2,150/kW bought locational value or just capacity. (Source: Utility Dive, April 2 — September 2 entry)
⚡ Virtual Power Plants & Demand Flexibility
PG&E, Google and Rewiring America launched SHARE on September 3 — a virtual power plant funded entirely by a third party, enrolling roughly 21,000 already-installed flexible devices and deploying new hardware alongside them, aimed at relieving transmission constraints as early as this fall. The program combines two mechanisms a DERMS has to orchestrate simultaneously and that most programs treat as alternatives: aggregating devices that are already in the field — Tesla and Sunrun batteries, Renew Home thermostats, EVs — and deploying new Carrier battery-enabled heat pumps to eligible households in Santa Clara and Alameda counties. PG&E and Demand Side Analytics aggregate the combined fleet. The proof-of-concept runs through 2027 with initial findings due late 2026 or early 2027, and PG&E flags a path to extend the model to commercial, industrial and utility-scale applications.
The structural novelty is the funding. Google pays for incentives, technology and program delivery. That removes the cost-recovery question that dominates every other VPP proceeding in this digest and replaces it with a different one: what does the utility own at the end? PG&E owns the operational record, the enrollment relationships and the locational performance data. It does not own the assets, and it has not established a rate-recovery precedent for the next tranche. For a utility evaluating this template, that is the trade — speed and zero rate-base exposure now, against a repeatability question in 2028.
The targeting is the part worth copying regardless of who pays. SHARE is screened by where demand is growing, aimed at unlocking regional transmission capacity rather than generic system-wide peak shaving. That is the non-wires-alternative workflow Virginia’s HB434 now mandates by statute and that National Grid runs across 19 projects in Massachusetts, applied at scale by the largest utility in California. PG&E’s own framing is affordability and “speed-to-capacity” — deferring transmission and distribution build-out and putting “downward pressure on rates” — which is the same avoided-cost mechanism behind this series’ ~$66/kW-year DR benchmark and its 70 MW-defers-$150M illustration. (Source: PG&E Corporation via PR Newswire, September 3 — September 4 entry)
A Regulatory Assistance Project synthesis reframed the VPP as an affordability instrument rather than a reliability one, and supplied the cleanest published taxonomy yet of the three operating models a DERMS has to support. RMI’s Kevin Brehm framed VPPs as levers that cut demand when wholesale prices are high — reducing capacity and transmission costs — and, critically, that free load-interconnection headroom on distribution grids. That second mechanism is underused in DR business cases and is the one that connects demand-side programs to the large-load queue: headroom released by orchestrated DERs is headroom a new commercial or industrial customer can occupy without a wires upgrade.
The three models the panel distinguished are worth writing into a DERMS requirements document verbatim, because they impose different settlement architectures: utility-run VPPs (compensated directly or through third-party aggregators), market-participant VPPs bidding into energy, capacity or ancillary markets, and an emerging “bring your own capacity” hybrid. Sunrun’s Bronte Payne named the conditions for scale — robust and predictable compensation, revenue-grade metering, and customer protections including penalty-free opt-out. The revenue-grade metering point is now the fourth consecutive week this series has arrived at telemetry fidelity as the binding DERMS specification.
The operational detail is the payoff for procurement teams. National Grid is tapping DER aggregations across 19 non-wires-alternatives projects in Massachusetts with 7.2 MW active or committed, and it screens each constrained node for existing dispatchable resource before deciding between its ConnectedSolutions+ program and an RFP-based framework. That screening step is the operational form of what Virginia’s HB434 makes a statutory obligation. Xcel Colorado, meanwhile, is moving toward a “direct-participant” capacity model as 1 GW of distributed solar creates voltage and reverse-power-flow problems on a 7 GW system, running its state-mandated Dispatchable Distributed Generation solicitation and enrolling its 25 MW/year Active VPP so that the portfolio “looks like a four-hour battery” regardless of what devices are underneath. That last phrase is the product definition a capacity market can actually buy, and it is the clearest statement this series has recorded of what a DERMS is for. (Source: Utility Dive, August 19 — September 3 entry)
California’s Energy Commission reset the Demand Side Grid Support program for 2026, suspending one participation option, restricting another to incumbents, and paying VPP aggregators a 30% bonus — with four-hour storage now worth roughly $107.64/kW-year. For program year 2026 the CEC suspended Participation Option 1 outright and limited the Option 3 market-aware storage VPP track to aggregations that participated in October 2025, excepting bidirectional-EVSE aggregations. Aggregators receive an additional 30% bonus on capacity incentives. Option 4, the emergency load-flexibility VPP track, remains open to third parties, publicly owned utilities and CCAs.
The per-kW numbers are the reason this matters outside California: approximately $107.64/kW-year for four-hour, $96.88/kW for three-hour, and $80.74/kW for two-hour storage. All three sit above the ~$66/kW-year avoided-cost benchmark this series tracks — which is the point. A state program will pay a premium above pure resource-adequacy value to buy speed, certainty and locational flexibility, and the size of that premium (roughly 60% at the four-hour duration) is a defensible number to cite when a commission asks why a DR program should be priced above strict avoided cost.
The design signal is sharper than the price signal, though. Suspending an option and restricting another to prior-season participants, while simultaneously raising the price for orchestrated fleets, is a mature program deliberately steering scarce dollars away from standalone enrollment and toward dispatchable aggregations. Read alongside California’s parallel CPUC demand-response rulemaking, the message to a DERMS vendor is that the Option 3 / Option 4 structure is a compensation schema the platform must be able to bid into, measure against, and settle — not a program a utility administers around it. (Source: CALSSA, 2026 program guidelines — September 2 entry)
Movement on carried items — CalChoice went live; New Jersey is still on a blank page. CalChoice, the California Choice Energy Authority CCA joint-powers authority, reached its targeted end-of-August 2026 full launch of the Lunar Energy-backed DERMS and VPP that this series flagged in the August 28 digest as the cleanest available example of correct sequencing. The sequencing claim now has a completed instance behind it: CalChoice built the DERMS framework first and formed the VPP program on top of it, staging rollout from already-deployed behind-the-meter batteries, thermostats and EV chargers, with participation rules and reward levels set per member community rather than through one system-wide tariff. That last design choice is what makes a joint-powers structure workable and is directly relevant to any utility running differentiated programs across service territories.
New Jersey‘s VPP straw proposal (Docket QO26030099, comments closed August 17), issued under Governor Sherrill’s Executive Order No. 2 declaring a statewide energy emergency, would aggregate home batteries, smart thermostats and EV chargers across PSE&G, JCP&L, Atlantic City Electric and Rockland Electric, compensating participants through bill credits, discounts or direct payments. Program launch remains targeted for 2027. It is carried here only because it is the one venue in the three-jurisdiction compensation-design comparison this series has been running where the performance-versus-enrollment question is still genuinely open. (Sources: CalCCA, March 18; NJ BPU, July 27 — September 1 and August 31 entries)
The SEPA / NCCETC Q2 2026 policy roundup shows VPP orchestration hardening into standard regulatory practice, with three items that function as procurement documents rather than policy statements. Hawaii’s PUC (Docket 2026-0084) specified that its VPP grid-services program use remote dispatch “managed by a 3rd-party edge DERMS agent, selected competitively,” with a five-year minimum term and compensation that prioritizes performance — a DERMS RFP embedded inside a commission order, and this series’ standing example of retrofitting orchestration onto an already-saturated base. On the valuation side, DTE Electric proposed a two-year Residential Battery VPP Pilot paying $105/kW-year, capped at $504/year — a clean per-kW capacity price to set beside California’s $107.64 and the ~$66/kW-year avoided-cost benchmark, and notable for how closely the two state programs converge from different methodologies. Illinois’ Commerce Commission approved ComEd and Ameren Scheduled Dispatch VPP tariffs with performance-based annual incentives under the Clean and Reliable Grid Affordability Act, with ComEd’s BYOD program adding smart-thermostat and thermal-storage eligibility.
Maryland‘s PSC went furthest into the plumbing, standing up a Data Exchange Work Group and a third-party-accessible DER registry and directing BGE, Delmarva and Pepco to align interconnection tools with forthcoming VPP applications. A third-party-accessible registry is the least glamorous and most load-bearing item in this list: it is the integration substrate that determines whether an aggregator can identify and enroll eligible devices at all, and it is the piece most often deferred to implementation and then never built. Virginia enacted four VPP laws including a 150 MW Appalachian Power DER-aggregation pilot and a requirement that VPPs be modeled in IRPs. The convergence across five jurisdictions — competitive DERMS selection, performance-based compensation, DER registries, IRP modeling mandates — is now a template regulators expect rather than a set of experiments. (Source: DSIRE Insight / NCCETC-SEPA, July 23 — August 31 entry)
🔌 DERMS & Grid Integration Technology
Carried with new detail — the DERMS procurement bar is now written to SCADA-grade reliability, and the specification language is public. The Itron-authored T&D World essay appeared in last week’s digest; the September 4 entry foregrounds the procurement-specification content, which is the part with operational value. The claim is that in 2026 DERMS has moved “from the innovation budget into the capital plan” alongside ADMS, substation automation and AMI, and that utilities are consequently writing DERMS specifications that mirror SCADA infrastructure: round-the-clock reliability, defined response times, failover and redundancy. The evaluation question shifts from “can this work in a demonstration?” to “can it run as a production system?” — which is a different RFP, scored differently, with a different set of qualifying vendors.
The three pressures the essay names are the ones an IRP team can quantify: roughly 78 GW of coal retirements removing the upstream dispatchable flexibility distribution operators implicitly relied on; 200 GW of U.S. demand growth projected by 2030; and updated interconnection standards that have turned customer inverters, batteries and EV chargers into grid-interactive assets — operational variables rather than passive load. The architectural line remains the one to quote in a procurement document: DERMS must extend ADMS, sharing data and workflows, rather than standing up as a parallel silo, with AMI smart meters serving as the grid-edge sensing and control layer for feeder-level dispatch.
Two benchmarks are worth carrying into vendor conversations: Xcel’s Renewable Battery Connect aggregated more than 15 MW of customer battery flexibility in under six months toward Colorado’s first battery VPP, and Hawaiian Electric uses DERMS to manage reverse power flow under extreme rooftop-solar penetration. The first is a speed-to-enrollment comparable; the second is the failure mode that makes orchestration non-optional rather than value-added. (Source: T&D World, June 9 — September 4 entry; previously reported in the August 28 digest)
The reliability requirements MISO filed at FERC are, read as a specification, a DERMS requirements list with a regulatory deadline attached. Filed August 28 as part of MISO’s response to FERC’s June show-cause orders, the framework defines “large loads” above 50 MW and carves out “computational loads” — large loads with at least 25 MW of information-technology demand — so that data-center-specific rules can be targeted at resources that “may exhibit rapid and coordinated changes in demand, significant power-electronic behavior, and distinct responses to transmission system disturbances.”
Four requirement classes follow, and each maps to a capability utilities are already buying for distributed resources. Visibility: basic facility and modeling data plus real-time and day-ahead load forecasts. Phasor measurement units for high-resolution monitoring of computational loads. Ramp limits on how fast loads may swing. Ride-through minimums so loads do not disconnect during voltage and frequency events. Day-ahead forecasting and PMU-grade telemetry are the same analytics that quantify hosting capacity for a DER portfolio; the reliability mandate is therefore a de facto driver of DERMS investment, and the platform’s remit extends from thousands of small devices to a handful of enormous ones without changing the underlying architecture.
The load-forecast context is the reason MISO is moving now: it projects 1–2% annual demand growth through 2044 after roughly 0.5% flat growth from 2009 to 2024. MISO asked FERC to let the rules take effect December 4 and plans further filings by November 16 covering transmission products, cost-shift protections, and the treatment of generation serving “electrically proximate” large loads. That November 16 tranche is where the co-location economics get decided and belongs on the same watch list as the CAISO compliance filing due the same day. (Source: Utility Dive, September 1; MISO filing August 28 — September 3 entry)
🏗️ Data Centers & Large Load Growth
The Congressional Research Service issued a data-center-and-grid FAQ that converts the flexibility argument from a modeling exercise into a citable federal finding: NERC’s summer 2026 ERCOT peak forecast runs 3.7 GW — 4.6% — below summer 2025 because more data centers can be curtailed. Report R49326, issued September 1, is the authoritative federal framing of the thesis this series has been assembling, and that single number is the most useful thing in it. It is not a projection of what flexibility could deliver; it is a reduction already booked into a reliability assessment, attributable to load that grid operators can interrupt when needed to prevent emergencies. Any DR business case that has been arguing avoided capacity in the abstract now has a real-world instance to anchor on.
The report’s second contribution is the interconnection framing: flexible data centers may interconnect faster because they may not require new generation or transmission, and “spatial flexibility” — shifting workloads between locations — can lower costs for all customers. It catalogs the federal scaffolding this digest tracks (FERC RM26-4, the June 18 show-cause proceeding; DOE 202(c) backup-generation orders) alongside the state layer: new large-load rate classes, demand-response requirements, and voluntary developer commitments intended to shield residential customers from cost shifts.
The affordability counterweight is documented rather than argued: LBNL and NC State modeling ties data-center growth to retail price increases, notably in PJM’s Northern Virginia corridor. That is the pressure driving every large-load tariff in this digest, and it is why demand response and efficiency at roughly $20.70/MWh remain the cheapest available hedge. For DERMS and ADMS planning the report’s implicit conclusion is the operational one: day-ahead visibility and curtailment enforcement are becoming prerequisites for interconnection, not optional analytics — which is exactly what MISO’s filing above would codify. A federal reference document saying so is useful in a rate case in a way that a vendor white paper is not. (Source: Congressional Research Service R49326, September 1 — September 4 entry)
SEPA’s Q2 analysis maps where large-load tariffs are actually building in dispatchable flexibility: utilities proposed 25 new large-load tariffs and service rules in 2026, and roughly one-quarter of tracked tariffs and rules now include a concrete dispatchable-flexibility option. This is the tariff-design layer beneath the flexibility thesis, and it complements rather than repeats the 77-tariff database this series reported August 28. The database counted how many large-load tariffs exist; this analysis asks how many of them make the load curtailable, and finds curtailability migrating out of bespoke interconnection agreements and into standardized tariff offerings.
That migration is the structurally important part. A bespoke curtailment clause in one interconnection agreement is a negotiation. A dispatchable-flexibility option in a filed tariff is a product, available to every qualifying customer, priced, and reviewable by a commission — which effectively creates a new and very large demand-response resource class whose avoided-cost value competes directly against the residential and C&I programs in this series’ value stack. Utilities running both should expect to defend the relative pricing.
The affordability case rests on the Duke University Nicholas Institute finding this series has carried since June: a 1–2% reduction in data-center peak demand can lower electricity rates 0.5–2.8% while protecting reliability. Two additional Duke findings from the September 1 entry are worth pairing with it (deduplicated — the same research appears in both the September 1 and September 4 entries): the grid could absorb roughly 100 GW of new flexible data-center load with curtailment needed in fewer than 1% of annual hours across the 22 largest balancing areas, and the binding obstacle is not technology but the absence of common operating guidelines between hyperscalers and utilities for how and when flexibility is called. That gap is a measurement-and-controls gap, which is to say it is a DERMS problem with a governance wrapper. Enforcement of tariff-based curtailment obligations at 50 MW-plus single sites requires the same telemetry, forecasting and dispatch capability utilities are already buying to manage distributed customer DERs. (Sources: SEPA, Q2 2026; Utility Dive, June 26 — September 4 and September 1 entries)
PJM’s market monitor put a dollar figure on the bill those tariffs are trying to allocate: data-center load accounted for 9% — $10.48/MWh — of PJM wholesale power cost through July, via the capacity market alone, and $29.4 billion of capacity-revenue increases across the last four auctions. Monitoring Analytics’ first-half state-of-the-market report is the clearest quantification yet of what large-load growth is costing consumers, and the qualifier matters: the 9% figure comes from the capacity market before counting any data-center-driven increases in energy or transmission prices, so it is a floor rather than an estimate. The monitor’s own warning is that the $29.4 billion “will continue to grow until the issues associated with the addition of large data center loads are addressed.”
The backdrop makes the case sharper. Total wholesale power cost surged 46% year over year to $116.53/MWh, or $56.7 billion, in the first seven months, against $79.57/MWh ($38 billion) a year earlier — while average load rose only 1.7–2.6%. Cost growth is dominated by capacity and energy price escalation, not by volume. That decoupling is the strongest available argument for demand-side investment, because it means the marginal megawatt of avoided peak is worth far more than an average-cost calculation implies. Every megawatt of curtailable large load or dispatchable behind-the-meter flexibility that clears in place of new firm capacity is a direct offset against that $29.4 billion escalation, and the FERC statistical-sampling order below just made 4.9 GW of it eligible to clear. (Source: Utility Dive, August 27 — August 31 entry)
📋 Regulatory & Policy
FERC ordered PJM to accept statistical sampling in place of interval smart-meter data, finding PJM had “unjustly and unreasonably” locked out at least 4.9 GW of residential demand-response and VPP capacity. Ruling for Voltus and the Mission:data Coalition against PJM in Docket EL26-4 (order issued July 29), the Commission found it unjust and unreasonable for PJM to demand interval smart-meter data that aggregators mostly cannot obtain — utilities across PJM’s 13-state, 67-million-customer footprint have largely refused to release it, citing state data-privacy rules. The Commission agreed that under the status quo PJM was losing “at least 4.9 gigawatts of capacity … at a time when PJM is experiencing unprecedented load growth driven in part by hyperscale data center development, threatening reliability.”
This is the most consequential item in the digest for anyone whose business case depends on residential DR, and the reason is that it reframes a measurement rule as a capacity policy. 4.9 GW of residential flexibility priced out of the capacity market is 4.9 GW of peaker deferral and capacity-cost suppression left on the table — set that against Monitoring Analytics’ $29.4 billion and the scale is not marginal. Voltus expects the remedy to “unlock hundreds of megawatts of residential load over the next several years.”
Two operational consequences follow. First, statistical sampling becomes the bankable M&V pathway a DERMS must support — PJM already accepts it for non-smart-meter customers, and FERC found it “can be as or more accurate” than metered data across large numbers of homes. A platform that can only settle against interval meter data is now specified to a standard the Commission has declared unnecessary and exclusionary. Second, the near-term battleground is methodology, not hardware: FERC directed PJM to open a stakeholder proceeding with initial plans due within 45 days, and the sampling design that emerges will set the accuracy, confidence and audit requirements every residential aggregation in PJM settles against. The order also revives the resource-adequacy case CPower lost at FERC in 2024 for lack of evidence, which signals the Commission now treats aggregator data-blocking as a Federal Power Act matter rather than a state privacy question. Utilities in PJM that have declined to release interval data should read that as a change in exposure. (Source: Canary Media, August 13; FERC order July 29 — September 1 entry)
Virginia wrote non-wires alternatives into the planning rulebook. HB434 (Chapter 611), from the 2026 Regular Session, requires utilities to analyze deployment of non-wires alternatives — energy storage resources, customer-owned or customer-financed capacity, utility-owned or utility-contracted distributed generation, virtual power plants, flexible transmission and static synchronous compensators, synchronous condensers, and power-quality monitoring equipment — as part of planning, before committing to conventional wires builds.
The mechanism is what makes this more than a statement of preference. A screening obligation forces the avoided-cost and total-resource-cost comparison into the evidentiary record: if a 70 MW demand-response or storage portfolio can defer a wires project at lower total resource cost, the utility must show its work rather than default to steel in the ground. For DERMS procurement the second-order effect is more direct — a utility that must quantify what distributed resources can deliver at specific constrained locations needs hosting-capacity, forecasting and locational dispatch analytics it may not currently have, and the statute supplies the business-case justification for buying them. National Grid’s node-by-node screen across 19 Massachusetts NWA projects is the working example of what compliance looks like in practice. Virginia joins the set of states hardwiring DER evaluation into the front end of the capital-planning cycle rather than the back end of the approval process. (Source: Virginia LIS, HB434 Chapter 611 — September 2 entry)
Rhode Island’s PUC lowered the utility’s authorized return and opened a large-load and non-wires agenda in the same order. The August 21 decision in Rhode Island Energy’s distribution rate case (Docket 25-45-GE) denied RIE’s request to raise its allowed return on equity from 9.3% to 10%, and shifted the capital structure from 57%/43% to 52%/48% equity/debt. Both moves lower the utility’s revenue requirement — and therefore the avoided-cost benchmark that customer-side resources have to beat, which cuts against DER economics in a way advocates rarely acknowledge. The Commission also phased out line-extension allowances subsidizing new gas connections over three years and accepted a heat-pump rate study via a CLF/RIE settlement.
The forward-looking directive is the one to track: RIE must file an extra-large (20 MW or larger) electric-load tariff by December 31, 2026, designed to keep data centers and other large loads from degrading reliability, shifting costs to existing customers, or causing rate shock. That threshold is worth noting against the 50 MW convergence at the RTO layer — Rhode Island is setting its screen at 20 MW, TVA’s Capacity Commitment Charge attaches at 5 MW, and NERC’s computational-load criterion sits at 20 MW at ≥60 kV with ≥1 MW IT load. The divergence this series flagged last week is widening, not resolving. The Commission additionally signaled that upcoming RIE infrastructure requests should be tested against non-wires alternatives — battery storage, EV managed charging, vehicle-to-grid, demand response — and that smart-meter-enabled time-varying rates are the next docket. (Source: Green Energy Consumers Alliance, August 25; PUC decision August 21 — September 1 entry)
DOE issued its third Section 202(c) load-curtailment order of 2026, and separately, the generation-side use of the same statute is producing 65% less power than a year earlier. Order No. 202-26-41, issued to PJM on September 1 in response to a same-day application, directs specified units to run and authorizes PJM — in coordination with its Transmission Owners and Electric Distribution Companies — to direct backup generation at large loads to operate as a last resort before declaring an Energy Emergency Alert 3 or during one, across the 13-state footprint. It is the third 2026 use of the authority to push ≥50 MW single-delivery-point loads onto their own on-site generators. A pattern of recurring orders tied to heat-wave and storm peaks is now established enough to plan against, which argues for building curtailment and behind-the-meter dispatch into interconnection agreements and capacity forecasts rather than relying on the emergency backstop.
The inverse use of the same statute is the more instructive story, and it is the one to bring to an IRP proceeding. DOE has ordered 10 units at six power plants — five coal-fired — to run past their retirement dates via rolling 90-day Section 202(c) orders, unprecedented in the agency’s roughly 50-year history. Five of those plants produced just 1.5 million MWh in Q1 2026, down 65% from 4.3 million MWh a year earlier. Washington’s Centralia unit generated zero. Colorado’s Craig Unit 1 ran only from April 10 to 25. CenterPoint estimates $20.5 million and a 14-week outage to comply for a single 104 MW unit; NIPSCO’s Schahfer units are offline for repairs entirely. The mechanism is straightforward — owners deferred maintenance ahead of planned closures, so forcing the units to stay online triggers exactly the spending that destroys their value.
Set the two 202(c) applications side by side and they describe the same policy reaching for reliability from opposite directions. On the load side it can compel curtailment instantly, at no cost, from facilities that have made no preparation. On the generation side it is compelling operation from assets that are expensive, unreliable, and in two cases not generating at all. For an IRP practitioner the conclusion is that “keep the old plant running” is neither the cheap option nor the firm one, and the capacity-factor data is the evidence. (Sources: U.S. Department of Energy, September 1; Utility Dive, June 23 — September 3 and August 31 entries)
Standing context — FERC’s RM26-4 show-cause orders remain the scaffolding under all of the above. Six Section 206 orders issued June 18 to PJM, MISO, SPP, CAISO, ISO-NE and NYISO preliminarily found existing tariffs “unjust and unreasonable” for failing to address large-load and co-located-load integration, giving operators 60 days to file conforming revisions or justify the status quo. MISO’s August 28 filing above is the first substantive compliance response this series has been able to read in detail. The determination those proceedings will make — whether a hyperscale load interconnects as firm demand or as a flexible/curtailable resource — is the pivot for hosting capacity, DR value, and the load forecasts that drive every capacity procurement in this digest. (Source: FERC Docket RM26-4-000 — September 2 entry)
🔬 EPRI Research Spotlight
A direct check confirms no new major EPRI publication entered the research log this week — the eighth consecutive dry week. As established last week, this series now reports the gap as a finding about publication cadence rather than as an omission in coverage. FlexMosaic and the DCFlex expansion remain the most recent substantive releases, and a targeted verification search this week surfaced nothing newer than the July materials.
The item that matters most for this digest is unchanged in substance and increasingly consequential in timing. The 96-MW Aurora AI Factory in Manassas, Virginia — NVIDIA, Emerald AI, EPRI, Digital Realty and PJM Interconnection — is billed as the world’s first power-flexible AI factory and is intended to implement a reference design and certification standard for power-flexible AI infrastructure, with demonstration testing conducted through DCFlex. The test protocol is the reason to watch: precise, real-time responses to simulated grid stress events, including mimicked demand spikes during summer heatwaves and sudden drops in renewable generation. That is a controlled experiment in exactly the response characteristics that both MISO’s ride-through and ramp requirements and PJM’s ER26-3515 qualification standards are being written against. The coalition estimates the reference design, if adopted nationwide, could unlock roughly 100 GW of capacity on the existing system — the same figure Duke’s Nicholas Institute reaches from an entirely different method, which is a useful convergence.
The verification search this week reconfirmed the schedule: EPRI, NVIDIA, Emerald AI and partners plan to bring Aurora online in late 2026, having originally slated it for the first half of the year. The timing problem this series flagged last week now has a firmer shape. NERC’s computational-load standards are due at FERC by December 31, and Aurora’s commissioning data will very likely arrive after that filing. The standards will be written on modeling rather than measurement. That is an argument for conservative interconnection terms in the interim, not for waiting.
DCFlex’s own framing is worth carrying into large-load negotiations because it is the cleanest published decomposition of what a data center can actually offer: flexibility combines managed workloads, reduced AI plant energy consumption, and backup power. The program’s stated aim is to establish a common understanding of how much flexibility a large load can deliver, which is the precondition for shortening interconnection timelines and improving grid planning — and it is precisely the “common operating guidelines” gap that Duke’s researchers identified as the binding obstacle. The coalition spans Compass Datacenters, Constellation, Emerald AI, Google, National Grid, Nebius, NVIDIA, Oracle and PADO AI, across nine demonstration sites including Europe.
The FlexMosaic five-class taxonomy remains the valuation grammar to expect in binding flexibility agreements — Class A (infrequent extreme stress) through Class E (frequency stabilization), keyed to notification time, duration, frequency, depth and speed of response, with Classes D and E identified as unlocking the most system value. This week gives it a third concrete job. MISO’s filing proposes ramp limits and ride-through minimums as interconnection preconditions; DOE’s 202(c) order specifies a transition to backup generation as a last resort before EEA 3; SEPA finds a quarter of large-load tariffs now carrying a dispatchable-flexibility option. Those are three different products being written in three different venues, and none of them will price correctly against the others without a shared vocabulary for response characteristics. FlexMosaic is still the only published one precise enough to do it.
Verification item RESOLVED — the NERC computational-load standards draft is posted. This series has carried the missing draft as an open verification item for four weeks. A targeted search this week confirms that NERC Project 2026-02 posted three proposed foundational reliability standards — CLO-001-1 (Computational Load Interconnection, Studies, and Modeling Data), CLO-002-1 and CLO-003-1, plus limited conforming updates to the FAC standards — on August 19, 2026, opening a 30-day formal comment period closing September 18 with a 10-day ballot. The registered ballot pool closed September 3. The standards are drafted to align with the July 16 FERC order, with NERC Board adoption targeted for December 2026 and the FERC filing due December 31. The schedule risk this series flagged three weeks running has narrowed but not closed: an initial comment-and-ballot round completing September 18 leaves roughly ten weeks for additional ballots anticipated through the fall before a December board vote. Any utility with computational load in its territory should be reading CLO-001-1 now, because the modeling-data obligations it creates land on the utility, not on the data center.
Standing watch: Aurora AI Factory commissioning data and any further schedule slip; the CLO-001-1 comment record and fall ballot results; FlexMosaic uptake in the ER26-3515 and RM26-4 compliance records; and EPRI.com direct check again next run — ninth week.
🚩 Utility-Sector Relevance Flags
⚑ Demand-Side Capacity Now Has Three Funding Structures, and They Differ by More Than 3x
Topic: VPP Funding Models / Capital Planning / Cost Recovery Strategy
Relevance: In one week: DOE compelled 50 MW-plus curtailment in PJM at zero cost and zero compensation; Minnesota approved Xcel’s rate-based Capacity*Connect at $430M for 200 MW (~$2,150/kW) against Xcel’s own $624/kW Colorado VPP; and PG&E launched a 21,000-device VPP funded entirely by Google. Three ways to obtain demand-side capacity, three completely different balance-sheet consequences, and only one of them creates a cost-recovery precedent the utility can repeat. The rate-based option is under active attack for not having been competitively procured; the third-party-funded option buys speed and operational learning but leaves the repeatability question unanswered in 2028.
Action Signal: Engage — Before the next VPP or DR filing, run the funding-structure comparison explicitly rather than defaulting to rate base. If a corporate counterparty with load-growth exposure in your territory would fund a proof-of-concept, the PG&E/Google structure gets a program into the field without a contested cost-recovery docket. Document what the utility retains — enrollment relationships, locational performance data, dispatch experience — because that is the asset, and be candid internally that the second tranche will still need a funding answer.
⚑ FERC Just Made 4.9 GW of Residential DR Eligible, and the Gate Was an M&V Rule
Topic: Wholesale Market Access / Measurement & Verification / DERMS Specification
Relevance: FERC ruled for Voltus and Mission:data (Docket EL26-4) that PJM’s insistence on interval smart-meter data was unjust and unreasonable, finding PJM was losing at least 4.9 GW of DR/VPP capacity across a 13-state, 67-million-customer footprint — and that statistical sampling “can be as or more accurate” across large numbers of homes. Set that 4.9 GW against Monitoring Analytics’ $29.4 billion of data-center-driven capacity-revenue increases: the capacity was installed, in customers’ homes, and blocked by a measurement rule. PJM must open a stakeholder proceeding with initial plans due within 45 days.
Action Signal: Implement — Two things immediately. First, audit whether your DERMS or DR vendor can settle against statistical sampling as well as interval data; a platform specified only to interval-meter settlement is now built to a standard FERC has declared exclusionary, and that is a contract-amendment conversation to have before the PJM sampling methodology is set. Second, if your utility has been declining to release interval data to aggregators on state-privacy grounds, that position has shifted from a defensible state matter to a Federal Power Act exposure. Get ahead of it internally.
⚑ Interconnection Requirements Are Becoming DERMS Requirements, With a December 4 Date
Topic: Large-Load Interconnection / Telemetry & Forecasting / Platform Scope
Relevance: MISO’s August 28 FERC filing would require, as a precondition of interconnection for computational loads (≥25 MW IT demand): visibility with real-time and day-ahead load forecasts, PMU monitoring, ramp limits, and ride-through minimums — requested effective December 4, with a further filing due November 16 on transmission products, cost-shift protections and generation serving electrically proximate loads. Every one of those four capability classes is something a DERMS already does for distributed resources. The platform’s remit is extending from thousands of small devices to a handful of very large ones, and the driver is a reliability mandate rather than a program budget.
Action Signal: Implement — Write large-load telemetry and day-ahead forecasting into the DERMS scope now, in the same specification as distributed-device orchestration, rather than procuring a separate large-load monitoring system later. The requirements are converging and the budgets should too. If you are in MISO, the November 16 filing is the one that determines co-location economics — put it on the calendar alongside CAISO’s compliance filing the same day.
⚑ A Federal Document Now Says Flexibility Already Cut 3.7 GW Off a Peak Forecast
Topic: DR Business Case / Load Forecasting / Regulatory Citations
Relevance: CRS Report R49326 (September 1) reports NERC’s finding that the summer 2026 ERCOT peak forecast runs 3.7 GW (4.6%) lower than 2025 “because more data centers can be curtailed by grid operators.” This is not a projection of what flexibility could deliver — it is a reduction already booked into a reliability assessment. Every DR business case this series has tracked has had to argue avoided capacity in the abstract against a firm-capacity alternative with a concrete price tag. That asymmetry just narrowed, and the source is a nonpartisan federal reference document rather than an advocacy paper or a vendor study.
Action Signal: Implement — Put R49326 and the 3.7 GW figure into the standing DR and IRP evidence file, and cite it wherever a proceeding has been treating demand-side capacity as speculative relative to supply-side capacity. Pair it with Duke’s ~100 GW-absorbable / <1%-of-hours finding and the CRS interconnection-speed argument. A CRS report is citable in a commission proceeding in a way that a consultant deck is not, and it is the single strongest addition to the file this quarter.
⚑ Curtailability Is Migrating From Interconnection Agreements Into Filed Tariffs
Topic: Large-Load Tariff Design / DR Resource Classes / Internal Program Competition
Relevance: SEPA counts 25 new large-load tariffs and service rules proposed in 2026, with roughly one-quarter of tracked tariffs and rules now carrying a concrete dispatchable-flexibility option — a shift from bespoke, negotiated curtailment clauses toward standardized, commission-reviewable products. Duke’s finding that a 1–2% data-center peak cut lowers rates 0.5–2.8% is the affordability justification. The consequence most utilities have not planned for is internal: a dispatchable-flexibility tariff creates a very large new DR resource class whose avoided-cost value competes directly against the residential and C&I programs in the same portfolio, and a commission will eventually ask why they are priced differently.
Action Signal: Engage — Before filing a large-load flexibility tariff, reconcile its capacity value against the existing DR portfolio’s, and be prepared to explain the spread on grounds other than customer class. Reference points now available: California DSGS at $107.64/kW-year (4-hour storage), DTE’s proposed $105/kW-year residential battery pilot, and this series’ ~$66/kW-year avoided-cost benchmark. Two independent state programs converging within $3/kW-year is a defensible band; a large-load tariff priced far outside it needs a reason.
⚑ The Supply-Chain Order Just Changed the Risk Profile of Battery-Heavy DER Portfolios
Topic: Procurement Risk / Portfolio Composition / Program Design
Relevance: The August 26 executive order blocks certain foreign-made batteries, inverters and transformers from grid connection and gives the Energy Secretary discretion over already-installed equipment, layered on the February Foreign Entity of Concern rules. China accounts for roughly 80% of global lithium-ion supply-chain capacity and about 40% of 2025 U.S. inverter volumes. BNEF expects delays and cancellations. Every storage-anchored program in this digest — New Jersey’s, Xcel’s 200 MW, California’s DSGS tracks — carries that exposure. Demand response and efficiency at roughly $20.70/MWh carry none of it.
Action Signal: Watch — Do not restructure a portfolio on an executive order that has no implementation guidance yet, but do add a supply-chain sensitivity to any storage-dependent DER business case going to a commission this fall, and confirm whether pending procurements have change-in-law provisions that address grid-connection prohibition rather than only tariffs and tax credits. The retroactive-discretion clause over installed equipment is the unusual term; ask counsel about it specifically. Four U.S. battery plants online this year is a real but slow mitigant.
⚑ The Cheapest Capacity in the Portfolio May Be Interconnection Headroom, Not Megawatts
Topic: Non-Wires Alternatives / Hosting Capacity / DR Value Stack
Relevance: RMI’s framing at the RAP session — that VPPs free load-interconnection headroom on distribution grids — is the least-used mechanism in the DR value stack and the one most relevant to a utility with a large-load queue. Headroom released by orchestrated DERs is headroom a new commercial customer can occupy without a wires upgrade, which converts a demand-side program into a revenue-enabling investment rather than only a cost-avoidance one. National Grid operationalizes it across 19 NWA projects (7.2 MW active/committed) by screening each constrained node for existing dispatchable resource before choosing between ConnectedSolutions+ and an RFP. Virginia’s HB434 now makes that screen a statutory obligation.
Action Signal: Implement — Add released interconnection headroom as an explicit, separately-quantified benefit line in DR and VPP business cases, distinct from avoided generation capacity and avoided T&D. Utilities with load-growth queues can value it directly: a node where orchestration defers an upgrade and admits a paying new customer produces two benefit streams from one program dollar, and most cost-effectiveness tests currently count one.
⚑ Program Design Parameters Keep Outperforming Incentive Levels
Topic: DR Program Design / Storage Enrollment Economics / Compensation Structure
Relevance: California’s 2026 DSGS reset did four structural things — suspended Option 1, restricted Option 3 to October 2025 incumbents, added a 30% aggregator bonus, and kept Option 4 open to third parties, POUs and CCAs — and those choices will shape participation more than the headline $107.64/kW-year does. Hawaii specified a five-year minimum term and competitive selection of the edge DERMS agent. Illinois approved performance-based annual incentives. Maryland stood up a third-party-accessible DER registry. Sunrun named revenue-grade metering and penalty-free opt-out as the scaling conditions. This is the fifth consecutive week where term length, minimum size, eligibility windows and dispatch triggers do more work than the payment level.
Action Signal: Implement — Review the structural parameters of every DR and VPP program before revisiting incentive levels: minimum term (a one-year program structurally excludes storage regardless of price), participant minimum size, eligibility windows that lock out new entrants, dispatch trigger, and opt-out terms. Specify revenue-grade metering and a third-party-accessible DER registry in the program rules, not deferred to implementation — the registry is the item most often deferred and then never built, and without it an aggregator cannot identify eligible devices at all.
⚑ Large-Load Thresholds Keep Diverging — Rhode Island Adds a 20 MW Screen
Topic: Customer Segmentation / Compliance Screening / Tariff Alignment
Relevance: This series flagged threshold divergence last week as a standing screen; it widened rather than converged. Rhode Island’s PUC directed RIE to file a 20 MW-plus large-load tariff by December 31. MISO’s filing defines large loads above 50 MW but computational loads at ≥25 MW of IT demand. DOE’s 202(c) authority applies at ≥50 MW. NERC’s CLO criteria sit at ≥20 MW at ≥60 kV with ≥1 MW IT load. TVA’s Capacity Commitment Charge attaches at 5 MW. A single customer can fall inside four of these and outside two.
Action Signal: Implement — Maintain one reconciled large-load customer list screened against all thresholds simultaneously, sized by the lowest applicable one, and add a column for the MISO-style IT-demand sub-threshold, which is a different measurement than site peak and will not be in most customer records. The IT-demand carve-out is the newest and least-tracked screen; a customer at 40 MW site peak with 28 MW of IT load is a computational load under MISO’s definition and not a large load under PJM’s.
⚑ The Generation-Side 202(c) Record Is the Best Available Argument Against “Just Keep the Plant Running”
Topic: IRP Advocacy / Resource Adequacy / Reliability Economics
Relevance: DOE has ordered 10 units at six plants (five coal-fired) to run past retirement via rolling 90-day orders. Five of them produced 1.5 million MWh in Q1 2026, down 65% year over year. Centralia generated zero; Craig Unit 1 ran two weeks in April. CenterPoint faces $20.5 million and a 14-week outage to comply for one 104 MW unit; NIPSCO Schahfer is offline entirely. Deferred maintenance ahead of planned closure means compelled operation triggers precisely the spending that destroys the asset’s value — and the units are not delivering the firmness the orders assume.
Action Signal: Engage — Bring the capacity-factor data into any IRP or resource-adequacy proceeding where delayed retirement is offered as the reliability answer. The argument is not that keeping units online is bad policy; it is that the empirical record shows it is neither cheap nor firm, which changes the comparison against a demand-side portfolio. The $20.5M-for-104-MW figure is the most concrete per-megawatt compliance cost available.
📌 Sources
August 31, 2026 Entry
– New Jersey Board of Public Utilities — NJ Proposes “Virtual Power Plant” Plan to Save Ratepayers Millions Annually, Docket QO26030099 (July 27, 2026) (previously reported in the August 28 digest; carried here only as context for the compensation-design flag)
– DSIRE Insight (NCCETC / SEPA) — VPP and Supporting DER Policy Developments: Q2 2026 (July 23, 2026)
– Utility Dive — Data Center Load Made Up 9% of PJM Wholesale Costs So Far in 2026: Market Monitor (August 27, 2026)
– Utility Dive — Power Plants Under DOE Emergency Orders Are Producing Way Less Energy Than Before (June 23, 2026)
September 1, 2026 Entry
– Canary Media — Federal Ruling Hands Virtual Power Plants a Win in PJM (August 13, 2026; FERC order issued July 29, 2026, Docket EL26-4)
– California Community Choice Association (CalCCA) — CalChoice Sets Sights on Virtual Power Plant With New DERMS Program (March 18, 2026; full launch end of August 2026) (previously reported in the August 28 digest; carried here as launch confirmation)
– Green Energy Consumers Alliance — Rhode Island Public Utilities Commission Made Some Good Calls (August 25, 2026; PUC decision August 21, 2026, Docket 25-45-GE)
– Utility Dive — Data Centers Are Ready to Negotiate Flexibility for Speed (June 26, 2026)
September 2, 2026 Entry
– Utility Dive — Minnesota Approves Xcel’s Controversial Utility-Owned Virtual Power Plant (April 2, 2026; MN PUC decision)
– FERC — Interconnection of Large Loads to the Interstate Transmission System, Docket No. RM26-4-000 (show-cause orders issued June 18, 2026)
– California Solar & Storage Association (CALSSA) — Demand Side Grid Support Program, 2026 Guidelines
– Virginia LIS — HB434, 2026 Regular Session (Chapter 611, CHAP0611)
September 3, 2026 Entry
– U.S. Department of Energy — Federal Power Act Section 202(c): PJM Interconnection, L.L.C., Order No. 202-26-41 (issued September 1, 2026)
– Utility Dive — Large Loads Face Reliability Requirements Under MISO Proposal (September 1, 2026; MISO filing at FERC August 28, 2026)
– Utility Dive — Trump Grid Order Likely to Cause Energy Storage Delays, Cancellations: BloombergNEF (September 1, 2026; executive order effective August 26, 2026)
– Utility Dive — VPP Value Proposition Expands to Affordability, Reliability and Resilience (August 19, 2026; RAP webinar August 18, 2026)
September 4, 2026 Entry
– PG&E Corporation (PR Newswire) — PG&E, Rewiring America and Google Launch First-of-its-Kind Virtual Power Plant to Help Lower Costs for All Customers (September 3, 2026)
– Congressional Research Service — Data Centers and the Electricity Grid: Frequently Asked Questions, Report R49326 (September 1, 2026)
– SEPA — Stretching the Possibilities: Where Large-Load Tariffs Fit in the Future of Data Center Flexibility (Q2 2026)
– T&D World — Why DERMS Is Earning a Place Alongside SCADA (Nick Tumilowicz, Itron; June 9, 2026) (previously reported in the August 28 digest; carried here for the procurement-specification detail)
Verification Sources (targeted search, not from the daily research log)
– NERC — Standards, Compliance and Enforcement Bulletin, August 31, 2026 (Project 2026-02 comment period through September 18)
– Climate Solutions Legal Digest — NERC Launches Project 2026-02 to Address Reliability Risks from Computational Loads
– POWER Magazine — FERC Orders Mandatory NERC Reliability Standards for Data Center and Other Computational Loads
– EPRI — DCFlex Initiative Expands to Nine Demonstration Sites Across U.S. and Europe
– Renewable Energy World — EPRI Expands DCFlex Data Center Initiative to Nine Demo Sites Across U.S. and Europe
– POWER Magazine — PJM Widens Response to Data Center Load as Capacity Shortfalls Deepen (Docket ER26-3515)